Multi-layer beamforming in uplink/downlink massive MIMO Systems with multi-antenna users

Abstract

In this paper, considering a hybrid architecture at the base station (BS) in an m-MIMO system, multi-antenna users transmit/receive multiple symbols to/from the BS in the uplink/downlink transmissions. In the uplink transmission, we propose a hybrid analog/digital beamformer at the BS and digital precoders at users. The BS designs the analog beamformer via signal-to-leakage-plus-noise ratio (SLNR) maximization criterion. Then, in the baseband, each user designs its precoder maximizing its effective signal power. Finally, the received signal at the BS is digitally combined with a regularized zero-forcing (RZF) filter. For the downlink transmission, we also design the hybrid beamformer at the BS and digital reception filters at users. In such a configuration, in order to design the hybrid beamformer, the digital precoders and the reception filters, we need much less channel knowledge compared to the large dimensional original channel matrices. Besides, as we show in the simulations, in spatially correlated channels, with smaller number of RF chains compared to the number of antennas, the performance of the proposed approaches would achieve the performance of the fully digital beamformer in terms of sum spectral efficiency (SE) and bit error rate (BER). Moreover, applying beamforming techniques at multi-antenna users further improves the overall network throughput.